12 |
|
|
13 |
use Data::Dumper; |
use Data::Dumper; |
14 |
|
|
15 |
|
#my $LOOKUP_REGEX = '\[[^\[\]]+\]'; |
16 |
|
#my $LOOKUP_REGEX_SAVE = '\[([^\[\]]+)\]'; |
17 |
|
my $LOOKUP_REGEX = 'lookup{[^\{\}]+}'; |
18 |
|
my $LOOKUP_REGEX_SAVE = 'lookup{([^\{\}]+)}'; |
19 |
|
|
20 |
=head1 NAME |
=head1 NAME |
21 |
|
|
22 |
WebPAC - base class for WebPAC |
WebPAC - base class for WebPAC |
188 |
|
|
189 |
push @{$self->{'data'}->{$mfn}->{$k}}, $val; |
push @{$self->{'data'}->{$mfn}->{$k}}, $val; |
190 |
} |
} |
191 |
|
} else { |
192 |
|
push @{$self->{'data'}->{$mfn}->{'000'}}, $mfn; |
193 |
} |
} |
194 |
|
|
195 |
} |
} |
198 |
my $rec = $self->{'data'}->{$mfn}; |
my $rec = $self->{'data'}->{$mfn}; |
199 |
$self->create_lookup($rec, @{$arg->{'lookup'}}); |
$self->create_lookup($rec, @{$arg->{'lookup'}}); |
200 |
|
|
201 |
|
$self->progress_bar($mfn,$maxmfn); |
202 |
|
|
203 |
} |
} |
204 |
|
|
205 |
$self->{'current_mfn'} = 1; |
$self->{'current_mfn'} = 1; |
206 |
|
$self->{'last_pcnt'} = 0; |
207 |
|
|
208 |
# store max mfn and return it. |
# store max mfn and return it. |
209 |
return $self->{'max_mfn'} = $maxmfn; |
return $self->{'max_mfn'} = $maxmfn; |
227 |
|
|
228 |
if ($mfn > $self->{'max_mfn'}) { |
if ($mfn > $self->{'max_mfn'}) { |
229 |
$self->{'current_mfn'} = $self->{'max_mfn'}; |
$self->{'current_mfn'} = $self->{'max_mfn'}; |
230 |
|
$log->debug("at EOF"); |
231 |
return; |
return; |
232 |
} |
} |
233 |
|
|
234 |
|
$self->progress_bar($mfn,$self->{'max_mfn'}); |
235 |
|
|
236 |
return $self->{'data'}->{$mfn}; |
return $self->{'data'}->{$mfn}; |
237 |
} |
} |
238 |
|
|
239 |
|
=head2 progress_bar |
240 |
|
|
241 |
|
Draw progress bar on STDERR. |
242 |
|
|
243 |
|
$webpac->progress_bar($current, $max); |
244 |
|
|
245 |
|
=cut |
246 |
|
|
247 |
|
sub progress_bar { |
248 |
|
my $self = shift; |
249 |
|
|
250 |
|
my ($curr,$max) = @_; |
251 |
|
|
252 |
|
my $log = $self->_get_logger(); |
253 |
|
|
254 |
|
$log->logconfess("no current value!") if (! $curr); |
255 |
|
$log->logconfess("no maximum value!") if (! $max); |
256 |
|
|
257 |
|
if ($curr > $max) { |
258 |
|
$max = $curr; |
259 |
|
$log->debug("overflow to $curr"); |
260 |
|
} |
261 |
|
|
262 |
|
$self->{'last_pcnt'} ||= 1; |
263 |
|
|
264 |
|
$self->{'last_pcnt'} = $curr if ($curr < $self->{'last_pcnt'}); |
265 |
|
|
266 |
|
my $p = int($curr * 100 / $max); |
267 |
|
if ($p != $self->{'last_pcnt'}) { |
268 |
|
printf STDERR ("%5d / %5d [%-51s] %-2d %% \r",$curr,$max,"=" x ($p/2).">", $p ); |
269 |
|
$self->{'last_pcnt'} = $p; |
270 |
|
} |
271 |
|
} |
272 |
|
|
273 |
=head2 open_import_xml |
=head2 open_import_xml |
274 |
|
|
275 |
Read file from C<import_xml/> directory and parse it. |
Read file from C<import_xml/> directory and parse it. |
293 |
|
|
294 |
$self->{'tag'} = $type2tag{$type_base}; |
$self->{'tag'} = $type2tag{$type_base}; |
295 |
|
|
296 |
$log->debug("using type '",$self->{'type'},"' tag <",$self->{'tag'},">") if ($self->{'debug'}); |
$log->info("using type '",$self->{'type'},"' tag <",$self->{'tag'},">"); |
297 |
|
|
298 |
my $f = "./import_xml/".$self->{'type'}.".xml"; |
my $f = "./import_xml/".$self->{'type'}.".xml"; |
299 |
$log->logconfess("import_xml file '$f' doesn't exist!") if (! -e "$f"); |
$log->logconfess("import_xml file '$f' doesn't exist!") if (! -e "$f"); |
300 |
|
|
301 |
$log->debug("reading '$f'") if ($self->{'debug'}); |
$log->info("reading '$f'"); |
302 |
|
|
303 |
|
$self->{'import_xml_file'} = $f; |
304 |
|
|
305 |
$self->{'import_xml'} = XMLin($f, |
$self->{'import_xml'} = XMLin($f, |
306 |
ForceArray => [ $self->{'tag'}, 'config', 'format' ], |
ForceArray => [ $self->{'tag'}, 'config', 'format' ], |
|
ForceContent => 1 |
|
307 |
); |
); |
308 |
|
|
309 |
|
$log->debug("import xml is ",sub { Dumper($self->{'import_xml'}) }); |
310 |
|
|
311 |
} |
} |
312 |
|
|
313 |
=head2 create_lookup |
=head2 create_lookup |
334 |
my $key = $self->fill_in($rec,$i->{'key'}); |
my $key = $self->fill_in($rec,$i->{'key'}); |
335 |
my @val = $self->fill_in($rec,$i->{'val'}); |
my @val = $self->fill_in($rec,$i->{'val'}); |
336 |
if ($key && @val && eval $eval) { |
if ($key && @val && eval $eval) { |
337 |
|
$log->debug("stored $key = ",sub { join(" | ",@val) }); |
338 |
push @{$self->{'lookup'}->{$key}}, @val; |
push @{$self->{'lookup'}->{$key}}, @val; |
339 |
} |
} |
340 |
} else { |
} else { |
341 |
my $key = $self->fill_in($rec,$i->{'key'}); |
my $key = $self->fill_in($rec,$i->{'key'}); |
342 |
my @val = $self->fill_in($rec,$i->{'val'}); |
my @val = $self->fill_in($rec,$i->{'val'}); |
343 |
if ($key && @val) { |
if ($key && @val) { |
344 |
|
$log->debug("stored $key = ",sub { join(" | ",@val) }); |
345 |
push @{$self->{'lookup'}->{$key}}, @val; |
push @{$self->{'lookup'}->{$key}}, @val; |
346 |
} |
} |
347 |
} |
} |
374 |
|
|
375 |
if ($$rec->{$f}) { |
if ($$rec->{$f}) { |
376 |
return '' if (! $$rec->{$f}->[$i]); |
return '' if (! $$rec->{$f}->[$i]); |
377 |
|
no strict 'refs'; |
378 |
if ($sf && $$rec->{$f}->[$i]->{$sf}) { |
if ($sf && $$rec->{$f}->[$i]->{$sf}) { |
379 |
$$found++ if (defined($$found)); |
$$found++ if (defined($$found)); |
380 |
return $$rec->{$f}->[$i]->{$sf}; |
return $$rec->{$f}->[$i]->{$sf}; |
415 |
This function B<does not> perform parsing of format to inteligenty skip |
This function B<does not> perform parsing of format to inteligenty skip |
416 |
delimiters before fields which aren't used. |
delimiters before fields which aren't used. |
417 |
|
|
418 |
|
This method will automatically decode UTF-8 string to local code page |
419 |
|
if needed. |
420 |
|
|
421 |
=cut |
=cut |
422 |
|
|
423 |
sub fill_in { |
sub fill_in { |
433 |
# FIXME remove for speedup? |
# FIXME remove for speedup? |
434 |
$log->logconfess("need HASH as first argument!") if ($rec !~ /HASH/o); |
$log->logconfess("need HASH as first argument!") if ($rec !~ /HASH/o); |
435 |
|
|
436 |
|
if (utf8::is_utf8($format)) { |
437 |
|
$format = $self->_x($format); |
438 |
|
} |
439 |
|
|
440 |
my $found = 0; |
my $found = 0; |
441 |
|
|
442 |
my $eval_code; |
my $eval_code; |
444 |
$eval_code = $1 if ($format =~ s/^eval{([^}]+)}//s); |
$eval_code = $1 if ($format =~ s/^eval{([^}]+)}//s); |
445 |
|
|
446 |
# do actual replacement of placeholders |
# do actual replacement of placeholders |
447 |
$format =~ s/v(\d+)(?:\^(\w))*/$self->get_data(\$rec,$1,$2,$i,\$found)/ges; |
$format =~ s/v(\d+)(?:\^(\w))?/$self->get_data(\$rec,$1,$2,$i,\$found)/ges; |
448 |
|
|
449 |
if ($found) { |
if ($found) { |
450 |
|
$log->debug("format: $format"); |
451 |
if ($eval_code) { |
if ($eval_code) { |
452 |
my $eval = $self->fill_in($rec,$eval_code,$i); |
my $eval = $self->fill_in($rec,$eval_code,$i); |
453 |
return if (! $self->_eval($eval)); |
return if (! $self->_eval($eval)); |
454 |
} |
} |
455 |
# do we have lookups? |
# do we have lookups? |
456 |
$log->debug("test format '$format' for lookups"); |
if ($format =~ /$LOOKUP_REGEX/o) { |
457 |
if ($format =~ /\[[^\[\]]+\]/o) { |
$log->debug("format '$format' has lookup"); |
458 |
return $self->lookup($format); |
return $self->lookup($format); |
459 |
} else { |
} else { |
460 |
return $format; |
return $format; |
481 |
|
|
482 |
my $tmp = shift || $log->logconfess("need format"); |
my $tmp = shift || $log->logconfess("need format"); |
483 |
|
|
484 |
if ($tmp =~ /\[[^\[\]]+\]/o) { |
if ($tmp =~ /$LOOKUP_REGEX/o) { |
485 |
my @in = ( $tmp ); |
my @in = ( $tmp ); |
486 |
|
|
487 |
$log->debug("lookup for: ",$tmp); |
$log->debug("lookup for: ",$tmp); |
488 |
|
|
489 |
my @out; |
my @out; |
490 |
while (my $f = shift @in) { |
while (my $f = shift @in) { |
491 |
if ($f =~ /\[([^\[\]]+)\]/) { |
if ($f =~ /$LOOKUP_REGEX_SAVE/o) { |
492 |
my $k = $1; |
my $k = $1; |
493 |
if ($self->{'lookup'}->{$k}) { |
if ($self->{'lookup'}->{$k}) { |
494 |
foreach my $nv (@{$self->{'lookup'}->{$k}}) { |
foreach my $nv (@{$self->{'lookup'}->{$k}}) { |
495 |
my $tmp2 = $f; |
my $tmp2 = $f; |
496 |
$tmp2 =~ s/\[$k\]/$nv/g; |
$tmp2 =~ s/lookup{$k}/$nv/g; |
497 |
push @in, $tmp2; |
push @in, $tmp2; |
498 |
} |
} |
499 |
} else { |
} else { |
503 |
push @out, $f; |
push @out, $f; |
504 |
} |
} |
505 |
} |
} |
506 |
|
$log->logconfess("return is array and it's not expected!") unless wantarray; |
507 |
return @out; |
return @out; |
508 |
} else { |
} else { |
509 |
return $tmp; |
return $tmp; |
534 |
|
|
535 |
$i = 0 if (! $i); |
$i = 0 if (! $i); |
536 |
|
|
537 |
my $format = $self->{'utf2cp'}->convert($format_utf8) || $log->logconfess("can't convert '$format_utf8' from UTF-8 to ",$self->{'code_page'}); |
my $format = $self->_x($format_utf8) || $log->logconfess("can't convert '$format_utf8' from UTF-8 to ",$self->{'code_page'}); |
538 |
|
|
539 |
my @out; |
my @out; |
540 |
|
|
541 |
|
$log->debug("format: $format"); |
542 |
|
|
543 |
my $eval_code; |
my $eval_code; |
544 |
# remove eval{...} from beginning |
# remove eval{...} from beginning |
545 |
$eval_code = $1 if ($format =~ s/^eval{([^}]+)}//s); |
$eval_code = $1 if ($format =~ s/^eval{([^}]+)}//s); |
547 |
my $prefix; |
my $prefix; |
548 |
my $all_found=0; |
my $all_found=0; |
549 |
|
|
550 |
while ($format =~ s/^(.*?)v(\d+)(?:\^(\w))*//s) { |
while ($format =~ s/^(.*?)v(\d+)(?:\^(\w))?//s) { |
551 |
|
|
552 |
my $del = $1 || ''; |
my $del = $1 || ''; |
553 |
$prefix ||= $del if ($all_found == 0); |
$prefix ||= $del if ($all_found == 0); |
564 |
|
|
565 |
return if (! $all_found); |
return if (! $all_found); |
566 |
|
|
567 |
my $out = join('',@out) . $format; |
my $out = join('',@out); |
568 |
|
|
569 |
|
if ($out) { |
570 |
|
# add rest of format (suffix) |
571 |
|
$out .= $format; |
572 |
|
|
573 |
# add prefix if not there |
# add prefix if not there |
574 |
$out = $prefix . $out if ($out !~ m/^\Q$prefix\E/); |
$out = $prefix . $out if ($out !~ m/^\Q$prefix\E/); |
575 |
|
|
576 |
|
$log->debug("result: $out"); |
577 |
|
} |
578 |
|
|
579 |
if ($eval_code) { |
if ($eval_code) { |
580 |
my $eval = $self->fill_in($rec,$eval_code,$i); |
my $eval = $self->fill_in($rec,$eval_code,$i); |
581 |
$log->debug("about to eval ",$eval," [$out]"); |
$log->debug("about to eval{",$eval,"} format: $out"); |
582 |
return if (! $self->_eval($eval)); |
return if (! $self->_eval($eval)); |
583 |
} |
} |
584 |
|
|
610 |
push @arr, $v; |
push @arr, $v; |
611 |
} |
} |
612 |
|
|
613 |
|
$log->debug("format '$format_utf8' returned ",--$i," elements: ", sub { join(" | ",@arr) }) if (@arr); |
614 |
|
|
615 |
return @arr; |
return @arr; |
616 |
} |
} |
617 |
|
|
618 |
|
=head2 fill_in_to_arr |
619 |
|
|
620 |
|
Similar to C<fill_in>, but returns array of all repeatable fields. Usable |
621 |
|
for fields which have lookups, so they shouldn't be parsed but rather |
622 |
|
C<fill_id>ed. |
623 |
|
|
624 |
|
my @arr = $webpac->fill_in_to_arr($rec,'[v900];;[v250^a]'); |
625 |
|
|
626 |
|
=cut |
627 |
|
|
628 |
|
sub fill_in_to_arr { |
629 |
|
my $self = shift; |
630 |
|
|
631 |
|
my ($rec, $format_utf8) = @_; |
632 |
|
|
633 |
|
my $log = $self->_get_logger(); |
634 |
|
|
635 |
|
$log->logconfess("need HASH as first argument!") if ($rec !~ /HASH/o); |
636 |
|
return if (! $format_utf8); |
637 |
|
|
638 |
|
my $i = 0; |
639 |
|
my @arr; |
640 |
|
|
641 |
|
while (my @v = $self->fill_in($rec,$format_utf8,$i++)) { |
642 |
|
push @arr, @v; |
643 |
|
} |
644 |
|
|
645 |
|
$log->debug("format '$format_utf8' returned ",--$i," elements: ", sub { join(" | ",@arr) }) if (@arr); |
646 |
|
|
647 |
|
return @arr; |
648 |
|
} |
649 |
|
|
650 |
|
|
651 |
=head2 data_structure |
=head2 data_structure |
652 |
|
|
653 |
Create in-memory data structure which represents layout from C<import_xml>. |
Create in-memory data structure which represents layout from C<import_xml>. |
655 |
|
|
656 |
my @ds = $webpac->data_structure($rec); |
my @ds = $webpac->data_structure($rec); |
657 |
|
|
658 |
|
This method will also set C<$webpac->{'currnet_filename'}> if there is |
659 |
|
<filename> tag in C<import_xml>. |
660 |
|
|
661 |
=cut |
=cut |
662 |
|
|
663 |
sub data_structure { |
sub data_structure { |
668 |
my $rec = shift; |
my $rec = shift; |
669 |
$log->logconfess("need HASH as first argument!") if ($rec !~ /HASH/o); |
$log->logconfess("need HASH as first argument!") if ($rec !~ /HASH/o); |
670 |
|
|
671 |
|
undef $self->{'currnet_filename'}; |
672 |
|
|
673 |
my @sorted_tags; |
my @sorted_tags; |
674 |
if ($self->{tags_by_order}) { |
if ($self->{tags_by_order}) { |
675 |
@sorted_tags = @{$self->{tags_by_order}}; |
@sorted_tags = @{$self->{tags_by_order}}; |
680 |
|
|
681 |
my @ds; |
my @ds; |
682 |
|
|
683 |
|
$log->debug("tags: ",sub { join(", ",@sorted_tags) }); |
684 |
|
|
685 |
foreach my $field (@sorted_tags) { |
foreach my $field (@sorted_tags) { |
686 |
|
|
687 |
my $row; |
my $row; |
689 |
#print "field $field [",$self->{'tag'},"] = ",Dumper($self->{'import_xml'}->{'indexer'}->{$field}->{$self->{'tag'}}); |
#print "field $field [",$self->{'tag'},"] = ",Dumper($self->{'import_xml'}->{'indexer'}->{$field}->{$self->{'tag'}}); |
690 |
|
|
691 |
foreach my $tag (@{$self->{'import_xml'}->{'indexer'}->{$field}->{$self->{'tag'}}}) { |
foreach my $tag (@{$self->{'import_xml'}->{'indexer'}->{$field}->{$self->{'tag'}}}) { |
692 |
my @v = $self->parse_to_arr($rec,$tag->{'content'}); |
my $format = $tag->{'value'} || $tag->{'content'}; |
693 |
|
|
694 |
|
$log->debug("format: $format"); |
695 |
|
|
696 |
|
my @v; |
697 |
|
if ($format =~ /$LOOKUP_REGEX/o) { |
698 |
|
@v = $self->fill_in_to_arr($rec,$format); |
699 |
|
} else { |
700 |
|
@v = $self->parse_to_arr($rec,$format); |
701 |
|
} |
702 |
next if (! @v); |
next if (! @v); |
703 |
|
|
704 |
|
# use format? |
705 |
|
if ($tag->{'format_name'}) { |
706 |
|
@v = map { $self->apply_format($tag->{'format_name'},$tag->{'format_delimiter'},$_) } @v; |
707 |
|
} |
708 |
|
|
709 |
# does tag have type? |
# does tag have type? |
710 |
if ($tag->{'type'}) { |
if ($tag->{'type'}) { |
711 |
push @{$row->{$tag->{'type'}}}, @v; |
push @{$row->{$tag->{'type'}}}, @v; |
713 |
push @{$row->{'display'}}, @v; |
push @{$row->{'display'}}, @v; |
714 |
push @{$row->{'swish'}}, @v; |
push @{$row->{'swish'}}, @v; |
715 |
} |
} |
716 |
|
|
717 |
|
if ($field eq 'filename') { |
718 |
|
$self->{'current_filename'} = join('',@v); |
719 |
|
$log->debug("filename: ",$self->{'current_filename'}); |
720 |
|
} |
721 |
|
|
722 |
} |
} |
723 |
|
|
724 |
if ($row) { |
if ($row) { |
725 |
$row->{'tag'} = $field; |
$row->{'tag'} = $field; |
726 |
|
|
727 |
|
# TODO: name_sigular, name_plural |
728 |
|
my $name = $self->{'import_xml'}->{'indexer'}->{$field}->{'name'}; |
729 |
|
$row->{'name'} = $name ? $self->_x($name) : $field; |
730 |
|
|
731 |
push @ds, $row; |
push @ds, $row; |
732 |
|
|
733 |
|
$log->debug("row $field: ",sub { Dumper($row) }); |
734 |
} |
} |
735 |
|
|
736 |
} |
} |
768 |
return $out; |
return $out; |
769 |
} |
} |
770 |
|
|
771 |
|
=head2 apply_format |
772 |
|
|
773 |
|
Apply format specified in tag with C<format_name="name"> and |
774 |
|
C<format_delimiter=";;">. |
775 |
|
|
776 |
|
my $text = $webpac->apply_format($format_name,$format_delimiter,$data); |
777 |
|
|
778 |
|
Formats can contain C<lookup{...}> if you need them. |
779 |
|
|
780 |
|
=cut |
781 |
|
|
782 |
|
sub apply_format { |
783 |
|
my $self = shift; |
784 |
|
|
785 |
|
my ($name,$delimiter,$data) = @_; |
786 |
|
|
787 |
|
my $log = $self->_get_logger(); |
788 |
|
|
789 |
|
if (! $self->{'import_xml'}->{'format'}->{$name}) { |
790 |
|
$log->warn("<format name=\"$name\"> is not defined in ",$self->{'import_xml_file'}); |
791 |
|
return $data; |
792 |
|
} |
793 |
|
|
794 |
|
$log->warn("no delimiter for format $name") if (! $delimiter); |
795 |
|
|
796 |
|
my $format = $self->_x($self->{'import_xml'}->{'format'}->{$name}->{'content'}) || $log->logdie("can't find format '$name'"); |
797 |
|
|
798 |
|
my @data = split(/\Q$delimiter\E/, $data); |
799 |
|
|
800 |
|
my $out = sprintf($format, @data); |
801 |
|
$log->debug("using format $name [$format] on $data to produce: $out"); |
802 |
|
|
803 |
|
if ($out =~ m/$LOOKUP_REGEX/o) { |
804 |
|
return $self->lookup($out); |
805 |
|
} else { |
806 |
|
return $out; |
807 |
|
} |
808 |
|
|
809 |
|
} |
810 |
|
|
811 |
|
|
812 |
# |
# |
813 |
# |
# |
814 |
# |
# |
861 |
return $va <=> $vb; |
return $va <=> $vb; |
862 |
} |
} |
863 |
|
|
864 |
|
=head2 _get_logger |
865 |
|
|
866 |
|
Get C<Log::Log4perl> object with a twist: domains are defined for each |
867 |
|
method |
868 |
|
|
869 |
|
my $log = $webpac->_get_logger(); |
870 |
|
|
871 |
|
=cut |
872 |
|
|
873 |
sub _get_logger { |
sub _get_logger { |
874 |
my $self = shift; |
my $self = shift; |
875 |
|
|
876 |
my @c = caller(1); |
my $name = (caller(1))[3] || caller; |
877 |
return get_logger($c[3]); |
return get_logger($name); |
878 |
|
} |
879 |
|
|
880 |
|
=head2 _x |
881 |
|
|
882 |
|
Convert string from UTF-8 to code page defined in C<import_xml>. |
883 |
|
|
884 |
|
my $text = $webpac->_x('utf8 text'); |
885 |
|
|
886 |
|
=cut |
887 |
|
|
888 |
|
sub _x { |
889 |
|
my $self = shift; |
890 |
|
my $utf8 = shift || return; |
891 |
|
|
892 |
|
return $self->{'utf2cp'}->convert($utf8) || |
893 |
|
$self->_get_logger()->logwarn("can't convert '$utf8'"); |
894 |
} |
} |
895 |
|
|
896 |
# |
# |